- Monday, March 23, 2015

dprive@jabber.ietf.org

Monday, March 23, 2015< ^ >

Dan York has set the subject to: DPRIVE at IETF91

Room Configuration

Room Occupants

GMT+0
[12:22:13] ilari.liusvaara joins the room
[20:06:16] jelte joins the room
[20:09:44] tjw.ietf joins the room
[20:16:28] shollenbeck joins the room
[20:16:33] yone joins the room
[20:16:49] Dan York 2 joins the room
[20:17:10] Dan York 2 has set the subject to: DPRIVE at IETF92
[20:17:12] Dan York joins the room
[20:17:39] C Peters joins the room
[20:17:44] Alexa Morris joins the room
[20:19:16] Dow Summers joins the room
[20:19:33] Jelte Jansen joins the room
[20:20:14] Jelte Jansen leaves the room
[20:20:43] Jelte Jansen joins the room
[20:20:59] Paul Selkirk joins the room
[20:22:12] Wouter Wijngaards joins the room
[20:22:32] <Dan York 2> We hear you remotely
[20:22:58] ajsaf@jabber.org joins the room
[20:23:08] Frederico A C Neves joins the room
[20:23:14] <Jelte Jansen> and a thank you from us remotes :)
[20:23:20] Jason Fesler joins the room
[20:23:30] <jelte> oh i'm here twice now
[20:23:34] Francis Dupont joins the room
[20:24:10] Alexa Morris leaves the room
[20:24:10] PaulWouters joins the room
[20:24:24] Ted.h joins the room
[20:24:32] Hugo Kobayashi joins the room
[20:24:41] <PaulWouters> note: the room is way too cold :P
[20:24:42] <Ted.h> If you would like something said at the mic, please say "MIC" at the beginning of your comment
[20:24:47] <jelte> bottom of the slide is white on white :p
[20:24:52] Suzanne joins the room
[20:25:13] <Ted.h> Warren is switching mics, to see if that works better.
[20:25:22] <Ted.h> Any agenda bash?
[20:25:28] <jelte> remote audio was already fine
[20:25:28] <Ted.h> Status update slide.
[20:25:37] Ralf Weber joins the room
[20:25:38] weiler joins the room
[20:26:02] <Ted.h> Plan slide:  decide on way forward
[20:26:28] <Ted.h> Possibly A, B, C, or N with the understanding that n-q will die.
[20:26:50] <Ted.h> End of administrivia
[20:26:57] <Ted.h> Allison coming up
[20:27:08] <Ted.h> "A touch of Eval" slide up
[20:27:48] <Ted.h> Now "Draft is posted" slide.
[20:27:59] <Ted.h> Asking for adoption.
[20:28:35] <Ted.h> Issues (1) Attacker model (emphasizing Type 1-A, pervasive passive monitor)
[20:28:53] wilma joins the room
[20:29:13] Jinmei Tatuya joins the room
[20:29:27] <Ted.h> System Model and Attackers slide (key on following slide)
[20:30:02] <Ted.h> Attacker is conceived of as a monitor.
[20:30:38] <Ted.h> Now on "Quick Key" slide
[20:31:07] <Ted.h> Describing mechanism parameters in templates
[20:32:27] <Ted.h> Which monitor are you depriving of the ability to see your traffic?  is one way to describe (What attacker are you thwarting?)
[20:33:22] Dan Wing joins the room
[20:33:48] <Ted.h> Issues 2 slide, "fill in TODOs" and ask for adoption.
[20:34:06] <Ted.h> Stephane at the mic
[20:34:40] Meetecho joins the room
[20:34:49] frodek joins the room
[20:35:22] Gabriel Montenegro joins the room
[20:36:21] <jelte> might be well known, but fair warning for jabber users: direct messages to people that are in the room through meetecho do not arrive
[20:36:26] <Ted.h> Aren't these from RFC 4949?
[20:37:13] <tjw.ietf> I thought from 6973
[20:37:33] KY Hong joins the room
[20:38:15] <Ted.h> That one cites 4949, so it may be both.
[20:38:38] <Ted.h> The problem statement document does not use this terminology; is that an issue?
[20:38:49] each joins the room
[20:39:06] <Ted.h> Allison replies that this in eval document, and we don't need to formalize the attack surface.
[20:39:12] Benno Overeinder joins the room
[20:39:41] <Ted.h> Fulll quote: "A foolish consistency is the hobgoblin of little minds, adored by little statesmen and philosophers and divines."
[20:39:52] <Ted.h> Ralph Waldo Emerson.
[20:40:05] <tjw.ietf> Thank you Ted. That was star worthy by itself!
[20:41:37] <Ted.h> Andrew gets the last word….there isn't any reason why this needs eleventy billion more spins; we should just get the work done and get it out.  We can spin later.
[20:41:48] <Ted.h> Phillip Hallam-Baker presenting "Private DNS"
[20:41:57] <Ted.h> Title slide
[20:42:23] <Ted.h> Remember to preface questions with "MIC", please.
[20:42:34] <Ted.h> Objectives slide.
[20:42:54] <Ted.h> Initially tackled a larger problem, re-scoped to the DNS.
[20:43:09] <Ted.h> Confidentiality and Traffic Analysis resistance.
[20:43:14] <Ted.h> are Privacy goals
[20:43:38] <Ted.h> Want to eliminate response spoofing, and choice of resolver.
[20:44:40] <Ted.h> Service should protect the resolver from resource exhaustion (not defeat DDOS mitigation) and protect third parties (amplification)
[20:44:47] <Ted.h> [Note:  front mic fixed now]
[20:44:53] <Ted.h> Constraints slight
[20:46:01] <Ted.h> Reflecting constraints from browser vendors in OCSP; he expects similar for private DNS.
[20:46:10] <Ted.h> EKR at the back mic
[20:46:45] <Ted.h> These are reasonable goals in terms of latency, but what made OCSP problematic was not this.
[20:47:22] <Ted.h> Phillip aims to avoid rat hole, and return to requirements.  100% usability, no additional latency, barring edge cases.
[20:48:11] Melinda joins the room
[20:48:14] <Ted.h> Architecture slide
[20:48:20] russ joins the room
[20:48:27] <Ted.h> Service connection establishment—consistent use of a single provider
[20:48:35] latourjacques joins the room
[20:48:48] <Ted.h> [SPLIT DNS roars in a corner, wounded, but not dead]
[20:49:46] <Ted.h> Phillip notes that longer bindings might be useful in certain circumstance (state sponsored actors preventing access to DNS).
[20:50:02] <ajsaf@jabber.org> [DNS-based CDNs wag their tails endearingly?]
[20:50:16] <Ted.h> Allison at the mic, notes work on how TOR's long lived context may be a parallel to examine.
[20:50:44] <Ted.h> John Dickinson:  The end user will have to choose a DNS provider and stick to it.
[20:50:56] <Ted.h> Phillip:  most users should want to.
[20:51:07] <Ted.h> John:  most users don't know anything about DNS and don't want to.
[20:51:31] <Ted.h> Phillip:  if you don't authenticate the server and you're telling the server where you're going, you're not going to have privacy.
[20:51:40] Olafur joins the room
[20:51:42] <Ted.h> But it is packaged, in our case with anti-virus.
[20:51:53] <Ted.h> Architecture II slide
[20:51:56] <tjw.ietf> "The Mom Support Paradox" should apply here
[20:51:58] <jelte> brrrr
[20:52:15] <Ted.h> Shows Resolution Protoco (UDP with light jacket, suitable for spring here in Dallas)
[20:53:01] <Ted.h> The idea here is that a single packet can hold all my DNS requests.
[20:53:21] <Ted.h> (contrasts with multiple queries feature now, and suggests "working around that bug")
[20:53:36] aregenscheid@darkdna.net joins the room
[20:53:40] aregenscheid@darkdna.net leaves the room
[20:53:50] a b joins the room
[20:54:01] aregenscheid@darkdna.net joins the room
[20:54:13] <Ted.h> EKR;  I want to go back to coffee shop scenario.
[20:54:44] <Ted.h> I agree with you that if you take any random resolver advertised in DHCP, that you will have no protection against active attackers.
[20:54:59] <Ted.h> Is there agreement that active attackers of this type are in scope?
[20:55:11] <Ted.h> Warren: we have not fully decided.
[20:55:34] <Ted.h> What had been discussed early was that some people would pick one and stick one, but others would not.
[20:55:53] <Ted.h> EKR:  this is making it hard to assess this.
[20:56:14] <Ted.h> Expectation is that folks who configure 8.8.8.8 now might do the tight binding here.
[20:56:35] <Ted.h> Phillip, there is no performance penalty here, slightly better
[20:56:50] <Ted.h> For some corner cases, a fall back to TLS may occur.
[20:57:18] <Ted.h> Applications:  Anonymous use (you will not authenticate yourself to the reolver)
[20:57:31] Olafur leaves the room
[20:57:37] <jelte> that 'workaround' might be an interesting approach even outside of dprive context
[20:57:50] <Ted.h> There is a quasi-anonymous use case; with a paid anti-virus you might see account details.
[20:57:51] russ leaves the room: Replaced by new connection
[20:57:52] Ning Kong joins the room
[20:57:55] russ joins the room
[20:58:11] <Ted.h> Enterprise would have to do split DNS, and they would need to drive OS adoption; there you need authentication of the user.
[20:58:17] <Ted.h> Complexity strategy slide
[20:58:40] <jelte> the padding thing wasn't received very well when i proposed it in the context of amplification attacks :p
[20:58:45] a b leaves the room
[20:58:58] <Ted.h> Now on Open Questions slide
[20:59:04] Erwin Karincic joins the room
[20:59:09] <Ted.h> Is an additional layer of crypto desirable?
[20:59:39] <Ted.h> Can add public key instead of kerberos ticket; this is meant to be simplest possible view (and wait on CFRG on ECC algorithins).
[20:59:54] Dan York leaves the room
[21:00:48] <jelte> "andrew pets the elephant"
[21:00:51] <Ted.h> Andrew Sullivan at the MIC:  we need to answer the question EKR asked to make progress.  But now asking:  if Architecture requires sending DNS packets over a different protocol.
[21:01:08] <Ted.h> Why is this not DNS two, which this WG is not chartered to do?
[21:01:27] marco@nl joins the room
[21:01:59] <Ted.h> Phillip:  the idea here is that changing the DNS client resolver protocol here should give us a moment to create an escape hatch from the current DNS world.
[21:03:10] <Ted.h> John asks: do you need standardization here, since you're shipping this inside a closed environment?  Answer:  we provide this now with proprietary protocols, but we'd like it be standards based.  I'd rather have my engineers doing interesting work, not supporting 20 platforms of proprietary stuff.
[21:03:40] <Ted.h> Paul Hoffman:  to answer EKR and ANDREW, the charter says "confidentiality between DNS clients and iterative resolvers".
[21:03:45] <Ted.h> Next up is Paul Hoffman
[21:03:59] <Ted.h> Paul has no slides.
[21:04:06] Olafur joins the room
[21:04:15] <Ted.h> He previously had three proposals, now zero.
[21:05:03] <Ted.h> He now believes a single way of doing it will work with middleboxes etc.
[21:05:17] Ning Kong leaves the room
[21:05:17] <Ted.h> Now co-author on the next draft, which describes a dual system.
[21:05:47] <Ted.h> draft-hzhwm-dprive-start-tls-for-dns now up, again with Paul.
[21:05:52] <Ted.h> Protocol overview slide
[21:05:59] <Ted.h> (changed since last IETF)
[21:06:23] Benno Overeinder leaves the room
[21:06:31] <Ted.h> TLS over TCP, DNS running in it.  Privacy comes from the properties of TLS.
[21:06:49] <Ted.h> Port-based initial test, then upgrade-based.
[21:07:33] <jelte> STARTTLS for dns?
[21:07:56] <Ted.h> We don't specify whether authenticated or opportunistic TLS
[21:08:09] <Ted.h> jelte: draft-hzhwm-dprive-start-tls-for-dns
[21:08:11] Frederico A C Neves leaves the room
[21:08:16] <Ted.h> Now on "Design Choices" slide
[21:08:38] <jelte> i'm a bit slow in connecting dots :)
[21:09:44] <Ted.h> Not mandated that a client try in a specific order; you can do upgrade first if history suggests that works.
[21:10:09] Ning Kong joins the room
[21:10:25] bortzmeyer joins the room
[21:11:30] <Ted.h> [At some point, you have to wonder whether running HTTP/2  with DNS uri targets doesn't end up doing this pretty completely; that gives you both the mechanism for asking and a standard response.  You can interleave, user server push, etc.]
[21:11:31] <tjw.ietf> Do we have a list of middleboxes which block TCP/53?
[21:12:30] <Ted.h> Eric Rescorla at the MIC:  If you the IP address of the resolver from DHCP doesn't help you much, if the binding is to an IP.
[21:13:21] <Ted.h> Another alternative is to give it a domain name as well as an IP address.
[21:13:22] Suzanne leaves the room
[21:13:22] ajsaf@jabber.org leaves the room
[21:13:49] <Ted.h> Paul notes that any decoration/identifier that you can use has the same property.
[21:14:47] <Ted.h> Eric notes that there is not much value of strong authentication if you would ever fallback to unencrypted DNS.
[21:15:46] <Ted.h> We would not be okay with indicia to distinguish this; we would be okay with a frob that said "always do this authenticated DNS, and do not fall back].
[21:15:53] <Ted.h> Changes since IETF 91 slide.
[21:16:02] <Ted.h> Implementation slide
[21:16:50] <Ted.h> Between stub and recursive, no reason to open and close TCP, but instead keep a long lived connection to amortize round trips.
[21:17:17] <Ted.h> Alison speaking, allay fears of millions of long lived connections.
[21:17:30] <Ted.h> This is a tuning matter, fast open gives you options here.
[21:18:33] <Ted.h> Peter Koch: since this is to inform the decision, the choosing the port and additional port discussion could be after adoption, if adopted?
[21:19:10] <Ted.h> What is the assumption of the end system using the authenticated DNS?  Multiplexed daemon?  App resolver connections?  Could be multiple per end system.
[21:20:38] tjw.ietf leaves the room
[21:20:38] <Ted.h> Alison RFC 5966bis has some guidance about this?
[21:21:00] <Ted.h> But what is the model?  Unified resolver or app-specific?
[21:21:09] <Ted.h> Paul: that model is not in the spec.
[21:21:14] <bortzmeyer> Ted.h: AFAIK, no (implementation issue, local decision)
[21:21:34] Linlin Zhou joins the room
[21:22:31] <Ted.h> Paolo: I'm not sure what problem this draft solves.  Your packets here are safe from your neighbors because of 802.11 link alyer encryption;  they are not safe from the resolvers.
[21:22:42] <weiler> That was Paul Wouters
[21:23:28] bortzmeyer is searching how to make an encrypted VPN with 8.8.8.8
[21:23:29] <Ted.h> John Heidemann:  if you use public dns, but no VPN, you have a risk.
[21:24:02] <weiler> bortzmeyer: +1
[21:24:14] <Ted.h> Wes: good job Paul for dodging the questions.  I have a problem seeing how any of this solves the bootstrapping authentication problem.
[21:24:26] <Ted.h> Many of the IPs are RFC 1918, so the IP doesn't work.
[21:24:37] <Ted.h> There's no easy anchor.
[21:24:47] <Ted.h> Maybe even go to DHCP and add another insert.
[21:25:14] tjw.ietf joins the room
[21:25:31] <Ted.h> Phillip:  UDP/TCP comment:  TCP is much worse for anycast.  I don't want to have any state on those servers.  Load balancers would have to keep lots of state, increasing my costs.
[21:25:44] tjw.ietf leaves the room
[21:26:10] <weiler> Previous comments from Wes Hardaker.  Currently Phillip Hallam Baker.
[21:26:13] <Ted.h> How will this get deployed?  linux package/browser?  It will get somewhere because browsers will implement at the application layer.
[21:26:29] <jelte> load balancers are worse than anycast i'd think. but probably also easier to fix
[21:27:14] <Ted.h> For most of the people who get deployment via browser, we don't need the opportunistic TLS.  So it shouldn't be the core of our design.
[21:27:49] Linlin Zhou leaves the room
[21:28:09] <Ted.h> Ralf Weber:  the DNS industry is currently running on cheap, low-cost system.  This will raise costs.
[21:28:35] <bortzmeyer> Ted.h: reading the bills at my employer, I wouldn't say "cheap"
[21:28:49] <Ted.h> One thing on middleboxes:  middleboxes inspect traffic and do intelligent things.  How do we get around that?
[21:29:08] <Ted.h> Alison:  We have that problem everywhere we're adding encryption.
[21:29:24] <Ted.h> Warren:  Using a different port means you are not triggering DNS ALGs.
[21:29:43] <Ted.h> Speaker:  why not encryption technology over UDP.  Is such available?
[21:30:14] <Ted.h> Alison:  the next document uses UDP.  We argue that amortizing the set-up works better, but  others disagree.
[21:30:19] <jelte> i don't think you can come up with anything that'll work with middleboxes that alter packets (apart from fail graciously)
[21:30:29] <Ted.h> Dan Wing:  I did write a DNS over DTLS spec before DPRIVE become a WG.
[21:30:57] <Ted.h> Now draft-wijngaards-dnsop-confidentialdn.  Glen Wiley speaking.
[21:31:04] <Ted.h> The slides are Wouter.
[21:31:35] <Ted.h> Believe this address all the main goals.  Has UDP and TCP choices, algorith agility.
[21:32:00] Paul Selkirk leaves the room
[21:32:09] <Ted.h> Offering opportunistic encryption, not adding much latency (one transaction per TTL)
[21:33:00] <Ted.h> Now on Updates 02-03 slide
[21:33:29] <Ted.h> Looking at key distribution.
[21:33:42] <Ted.h> Now Confidential DNS RRTYPE.
[21:34:30] <Ted.h> Has this been deployed?
[21:34:34] <Ted.h> Only in our dreams?
[21:34:37] <Ted.h> Did it work well there?
[21:35:27] <jelte> we can't hear whoever's talking now
[21:35:29] <Ted.h> Paul at the mic:  there's an issue with causing resolver meltdown with CPU ddos attack; TCP works, because of the return routability.
[21:36:02] <Ted.h> Retargeting them to other mics
[21:36:07] <jelte> thanks
[21:37:05] <Ted.h> Paul's not convinced that this uses much more CPU to get similar encryption, compared to other options.
[21:37:21] <Ted.h> Glen agrees that there are design trade-offs here.
[21:37:28] Linlin Zhou joins the room
[21:37:34] johnh joins the room
[21:37:39] <Ted.h> We're going to get new work here, one way or the other.
[21:37:59] <Ted.h> Philip Hallam-Baker:  you don't want to do public key work without at least return routability proven.
[21:38:32] <Ted.h> We've also got the bit that we won't forward your bits to the authoritative server unless you have also done a proof-of-work.
[21:38:40] <Ted.h> But there is a third issue:
[21:39:16] <Ted.h> The servers I use to do key negotiation and those for resolution are different, so that DDOS may hit new connections.
[21:39:16] frodek leaves the room
[21:39:28] <Ted.h> Ralf Weber:  This smells like DNS to me.
[21:40:14] <Ted.h> The attack vector described seems possible without prior authentication, and there may be some work done here to make the CPU issue tractable.
[21:40:27] <Ted.h> Lorenzo Colliti: how do you avoid HOL with TCP here?
[21:40:59] <Ted.h> Who will switch to TCP if you can do one query at a time?
[21:41:29] <Ted.h> Alison: 5966bis explains how to pipeline, we just don't do it now; we can turn it now.
[21:42:28] <Ted.h> Now on the "off to see the Wizard" slide, picking the path on the yellow brick road
[21:42:40] <Ted.h> [Step one, find house to drop on witch]
[21:42:46] Frederico A C Neves joins the room
[21:43:07] <Ted.h> One draft …or many? slide
[21:43:20] Erwin Karincic leaves the room
[21:43:28] <Ted.h> Whatever we decide now probably won't be final, if we find errors, we can pivot
[21:43:35] Dow Summers leaves the room
[21:43:45] <jelte> can we pick and choose parts of all proposals? :)
[21:44:00] <Ted.h> jelte:  once adopted, sure.
[21:44:03] Guangqing Deng joins the room
[21:44:12] Dow Summers joins the room
[21:44:15] <Ted.h> Hums coming
[21:44:42] <Ted.h> Prior to that:  which ones in the room have read the drafts and are ready?
[21:44:48] <Ted.h> Hum 1: choose an approach
[21:44:52] <Ted.h> Hum 2: choose multiple
[21:44:59] <Ted.h> Hum 3:  don't know enough yet.
[21:45:18] <Ted.h> Peter Koch: clarifying question
[21:45:42] <Ted.h> What does it mean "don't know enough" yet?  Warren; we make no decision now.
[21:46:26] <Ted.h> Adopting 0.
[21:46:48] <Ted.h> Hum1:  I would like to adopt a document
[21:46:55] <C Peters> hummmmm
[21:46:56] <Ted.h> Hum2:  I would like to adopt multiple
[21:47:01] <Jelte Jansen> hummmmm
[21:47:21] <Jelte Jansen> ok that is NOT as anonymous as a hum :p
[21:47:29] <weiler> three very similarly-volumed hums.
[21:47:31] <Ted.h> Hum 3:  Don't know enough yet
[21:47:46] <Ted.h> Very mixed; no clear result among these hums.
[21:47:49] shollenbeck leaves the room
[21:47:56] Gabriel Montenegro leaves the room
[21:48:16] <Ted.h> Sam:  is it worth doing a binary up down to see if any fall out?
[21:48:30] <Ted.h> Chairs:  yes, that might reduce the search space.
[21:49:02] Olafur leaves the room
[21:49:04] Jason Fesler leaves the room
[21:49:12] <Ted.h> Philip:  let's get the requirements from platform and browser folks.
[21:49:56] wilma leaves the room
[21:49:58] frodek joins the room
[21:50:31] <Ted.h> Christian:  implementation cost and deployment costs cause me to choose b:  least innovation.
[21:50:58] <Ted.h> Up down on each one.
[21:52:04] <Ted.h> Daniel Kahn Gilmoor: also support b; he is doing it now in unbound, though it needs performance enhancements.
[21:52:10] <Ted.h> It's available now.
[21:52:33] <Ted.h> Lars-Johann Liman; the cost of implementation is not the focus; it should be on the user.
[21:52:57] <Ted.h> private-dns yes
[21:53:00] <C Peters> hummmmm
[21:53:03] <Ted.h> private-dns no
[21:53:25] <jelte> oh god :)
[21:53:27] <Ted.h> draft-hzhwm yes
[21:53:27] <C Peters> hummmmm
[21:53:33] <Ted.h> draft-hahwm no
[21:53:33] <weiler> random reordering of questions…
[21:53:44] <Ted.h> draft-wijngaards yes
[21:53:47] <Ted.h> ditto no
[21:53:58] Hugo Kobayashi leaves the room
[21:54:04] Ning Kong leaves the room
[21:54:06] weiler leaves the room
[21:54:10] <Ted.h> 1st: more no than yes; 2: more yes than no; 3: balanced.
[21:54:12] aregenscheid@darkdna.net leaves the room
[21:54:13] yone leaves the room
[21:54:17] <Ted.h> Chairs:  taking it to the list.
[21:54:18] Ted.h leaves the room
[21:54:30] marco@nl leaves the room
[21:54:32] Melinda leaves the room
[21:54:33] Jinmei Tatuya leaves the room
[21:54:44] latourjacques leaves the room
[21:54:50] Francis Dupont leaves the room: Computer went to sleep
[21:55:14] frodek leaves the room
[21:55:21] <jelte> thanks scribe!
[21:55:35] Wouter Wijngaards leaves the room
[21:56:15] Jelte Jansen leaves the room
[21:56:20] Linlin Zhou leaves the room
[21:56:50] jelte leaves the room
[21:56:59] Ralf Weber leaves the room
[21:57:03] Frederico A C Neves leaves the room
[21:57:40] each leaves the room
[21:58:44] Dan Wing leaves the room
[22:00:43] Meetecho leaves the room
[22:02:02] PaulWouters leaves the room
[22:02:29] yone joins the room
[22:02:30] marco@nl joins the room
[22:02:49] marco@nl leaves the room: Replaced by new connection
[22:03:32] yone leaves the room
[22:03:34] Dow Summers leaves the room
[22:05:39] Olafur joins the room
[22:07:22] C Peters leaves the room
[22:08:03] Benno Overeinder joins the room
[22:08:35] russ leaves the room
[22:10:52] Benno Overeinder leaves the room
[22:11:02] Olafur leaves the room
[22:11:25] johnh leaves the room
[22:12:14] Jinmei Tatuya joins the room
[22:13:11] ajsaf@jabber.org joins the room
[22:13:17] KY Hong leaves the room
[22:14:15] ajsaf@jabber.org joins the room
[22:14:31] ajsaf@jabber.org leaves the room
[22:14:31] ilari.liusvaara leaves the room: offline
[22:14:32] ajsaf@jabber.org leaves the room
[22:15:05] Dan Wing joins the room
[22:15:07] Guangqing Deng leaves the room
[22:15:10] Jinmei Tatuya leaves the room
[22:15:18] bortzmeyer leaves the room
[22:16:01] Dan Wing leaves the room
[22:16:12] bortzmeyer joins the room
[22:19:33] Melinda joins the room
[22:20:03] latourjacques joins the room
[22:21:29] russ joins the room
[22:21:54] russ leaves the room
[22:23:05] Melinda leaves the room
[22:24:43] Ted.h joins the room
[22:27:15] johnh joins the room
[22:27:40] johnh leaves the room
[22:27:46] Ted.h leaves the room
[22:34:19] bortzmeyer leaves the room: Replaced by new connection
[22:34:30] bortzmeyer joins the room
[22:38:26] bortzmeyer leaves the room: Replaced by new connection
[22:38:37] bortzmeyer joins the room
[22:40:04] bortzmeyer leaves the room
[22:40:11] bortzmeyer joins the room
[22:42:45] Suzanne joins the room
[22:45:35] Suzanne leaves the room
[22:46:37] Frederico A C Neves joins the room
[22:51:32] PaulWouters joins the room
[22:52:28] Benno Overeinder joins the room
[22:52:34] Benno Overeinder leaves the room
[22:53:21] Frederico A C Neves leaves the room
[22:53:54] bortzmeyer leaves the room
[22:57:01] Olafur joins the room
[23:00:53] Ralf Weber joins the room
[23:08:48] Frederico A C Neves joins the room
[23:11:30] latourjacques leaves the room
[23:13:39] Olafur leaves the room
[23:27:23] Frederico A C Neves leaves the room
[23:29:53] Frederico A C Neves joins the room
[23:33:57] PaulWouters leaves the room
[23:49:30] Jason Fesler joins the room
[23:49:36] Jason Fesler leaves the room
[23:55:13] Ralf Weber leaves the room